arXiv subjects

Jian Yuan

Publications and source records attributed to Jian Yuan.

At least 37 records · Page 2Linked to original sources

arXiv2025

The Art of Tool Interface Design

We present an agentic framework, Thinker, which achieves state of art performance in challenging reasoning tasks for realistic customer service scenarios that involve complex business logic and human interactions via long horizons. On the $τ$-bench retail dataset, Thinker achieves 82.6\% success rate with GPT-4o (version 2024-06-01) (baseline: 68.3\%), and 81.9\% success rate with Llama-3.1 405B (baseline: 49.6\%), without any fine-tuning. Thinker effectively closes the gap in reasoning capabilities between the base models by introducing proper structure. The key features of the Thinker framework are: (1) State-Machine Augmented Generation (SMAG), which represents business logic as state machines and the LLM uses state machines as tools. (2) Delegation of tasks from the main reasoning loop to LLM-powered tools. (3) Adaptive context management. Our prompting-only solution achieves signficant gains, while still maintaining a standard agentic architecture with a ReAct style reasoning loop. The key is to innovate on the tool interface design, as exemplified by SMAG and the LLM-powered tools.

cs.AI↗

arXiv2025

WeightedKV: Attention Scores Weighted Key-Value Cache Merging for Large Language Models

Large Language Models (LLMs) use key-value (KV) cache to reduce redundant computation in autoregressive generation. However, the KV cache size increases linearly during generation, leading to excessive memory usage, especially for long texts. Most KV cache compression methods evict the unimportant KV pairs to maintain a fixed cache size, which leads to the permanent loss of tokens during generation. However, singular value decomposition shows that \textit{values} do not exhibit a strong low-rank property as \textit{keys} do, suggesting that information is distributed more evenly across \textit{values}, in contrast to its more redundant distribution within \textit{keys}. Therefore, methods that evict both \textit{keys} and \textit{values} risk losing crucial information and compromise context integrity, ultimately degrading the output quality. To address this problem, we propose WeightedKV, a novel, training-free approach that discards the \textit{keys} of less important tokens, while merging their \textit{values} into neighboring tokens via a convex combination weighted by their average attention scores. In this way, the retained \textit{keys} serve as anchors that guide the generation process, while the merged \textit{values} provide a rich contextual backdrop. We assess our method on four widely used language modeling datasets, demonstrating superior performance compared to all baseline methods, particularly with a lower budget ratio.

cs.CL↗

arXiv2025

LCSim: A Large-Scale Controllable Traffic Simulator

With the rapid growth of urban transportation and the continuous progress in autonomous driving, a demand for robust benchmarking autonomous driving algorithms has emerged, calling for accurate modeling of large-scale urban traffic scenarios with diverse vehicle driving styles. Traditional traffic simulators, such as SUMO, often depend on hand-crafted scenarios and rule-based models, where vehicle actions are limited to speed adjustment and lane changes, making it difficult for them to create realistic traffic environments. In recent years, real-world traffic scenario datasets have been developed alongside advancements in autonomous driving, facilitating the rise of data-driven simulators and learning-based simulation methods. However, current data-driven simulators are often restricted to replicating the traffic scenarios and driving styles within the datasets they rely on, limiting their ability to model multi-style driving behaviors observed in the real world. We propose \textit{LCSim}, a large-scale controllable traffic simulator. First, we define a unified data format for traffic scenarios and provide tools to construct them from multiple data sources, enabling large-scale traffic simulation. Furthermore, we integrate a diffusion-based vehicle motion planner into LCSim to facilitate realistic and diverse vehicle modeling. Under specific guidance, this allows for the creation of traffic scenarios that reflect various driving styles. Leveraging these features, LCSim can provide large-scale, realistic, and controllable virtual traffic environments. Codes and demos are available at https://tsinghua-fib-lab.github.io/LCSim.

cs.RO↗

arXiv2025

Noise Matters: Diffusion Model-based Urban Mobility Generation with Collaborative Noise Priors

With global urbanization, the focus on sustainable cities has largely grown, driving research into equity, resilience, and urban planning, which often relies on mobility data. The rise of web-based apps and mobile devices has provided valuable user data for mobility-related research. However, real-world mobility data is costly and raises privacy concerns. To protect privacy while retaining key features of real-world movement, the demand for synthetic data has steadily increased. Recent advances in diffusion models have shown great potential for mobility trajectory generation due to their ability to model randomness and uncertainty. However, existing approaches often directly apply identically distributed (i.i.d.) noise sampling from image generation techniques, which fail to account for the spatiotemporal correlations and social interactions that shape urban mobility patterns. In this paper, we propose CoDiffMob, a diffusion model for urban mobility generation with collaborative noise priors, we emphasize the critical role of noise in diffusion models for generating mobility data. By leveraging both individual movement characteristics and population-wide dynamics, we construct novel collaborative noise priors that provide richer and more informative guidance throughout the generation process. Extensive experiments demonstrate the superiority of our method, with generated data accurately capturing both individual preferences and collective patterns, achieving an improvement of over 32%. Furthermore, it can effectively replace web-derived mobility data to better support downstream applications, while safeguarding user privacy and fostering a more secure and ethical web. This highlights its tremendous potential for applications in sustainable city-related research. The code and data are available at https://github.com/tsinghua-fib-lab/CoDiffMob.

cs.LG↗

arXiv2025

Isolated Diffusion: Optimizing Multi-Concept Text-to-Image Generation Training-Freely with Isolated Diffusion Guidance

Large-scale text-to-image diffusion models have achieved great success in synthesizing high-quality and diverse images given target text prompts. Despite the revolutionary image generation ability, current state-of-the-art models still struggle to deal with multi-concept generation accurately in many cases. This phenomenon is known as ``concept bleeding" and displays as the unexpected overlapping or merging of various concepts. This paper presents a general approach for text-to-image diffusion models to address the mutual interference between different subjects and their attachments in complex scenes, pursuing better text-image consistency. The core idea is to isolate the synthesizing processes of different concepts. We propose to bind each attachment to corresponding subjects separately with split text prompts. Besides, we introduce a revision method to fix the concept bleeding problem in multi-subject synthesis. We first depend on pre-trained object detection and segmentation models to obtain the layouts of subjects. Then we isolate and resynthesize each subject individually with corresponding text prompts to avoid mutual interference. Overall, we achieve a training-free strategy, named Isolated Diffusion, to optimize multi-concept text-to-image synthesis. It is compatible with the latest Stable Diffusion XL (SDXL) and prior Stable Diffusion (SD) models. We compare our approach with alternative methods using a variety of multi-concept text prompts and demonstrate its effectiveness with clear advantages in text-image consistency and user study.

cs.CV↗

arXiv2024

Global Estimation of Building-Integrated Facade and Rooftop Photovoltaic Potential by Integrating 3D Building Footprint and Spatio-Temporal Datasets

This research tackles the challenges of estimating Building-Integrated Photovoltaics (BIPV) potential across various temporal and spatial scales, accounting for different geographical climates and urban morphology. We introduce a holistic methodology for evaluating BIPV potential, integrating 3D building footprint models with diverse meteorological data sources to account for dynamic shadow effects. The approach enables the assessment of PV potential on facades and rooftops at different levels-individual buildings, urban blocks, and cities globally. Through an analysis of 120 typical cities, we highlight the importance of 3D building forms, cityscape morphology, and geographic positioning in measuring BIPV potential at various levels. In particular, our simulation study reveals that among cities with optimal facade PV performance, the average ratio of facade PV potential to rooftop PV potential is approximately 68.2%. Additionally, approximately 17.5% of the analyzed samples demonstrate even higher facade PV potentials compared to rooftop installations. This finding underscores the strategic value of incorporating facade PV applications into urban sustainable energy systems.

cs.IR↗

arXiv2024

ODGEN: Domain-specific Object Detection Data Generation with Diffusion Models

Modern diffusion-based image generative models have made significant progress and become promising to enrich training data for the object detection task. However, the generation quality and the controllability for complex scenes containing multi-class objects and dense objects with occlusions remain limited. This paper presents ODGEN, a novel method to generate high-quality images conditioned on bounding boxes, thereby facilitating data synthesis for object detection. Given a domain-specific object detection dataset, we first fine-tune a pre-trained diffusion model on both cropped foreground objects and entire images to fit target distributions. Then we propose to control the diffusion model using synthesized visual prompts with spatial constraints and object-wise textual descriptions. ODGEN exhibits robustness in handling complex scenes and specific domains. Further, we design a dataset synthesis pipeline to evaluate ODGEN on 7 domain-specific benchmarks to demonstrate its effectiveness. Adding training data generated by ODGEN improves up to 25.3% mAP@.50:.95 with object detectors like YOLOv5 and YOLOv7, outperforming prior controllable generative methods. In addition, we design an evaluation protocol based on COCO-2014 to validate ODGEN in general domains and observe an advantage up to 5.6% in mAP@.50:.95 against existing methods.

cs.CV↗

arXiv2024

Giant domain wall anomalous Hall effect in an antiferromagnet

Generally, the dissipationless Hall effect in solids requires time-reversal symmetry breaking (TRSB), where TRSB induced by external magnetic field results in ordinary Hall effect, while TRSB caused by spontaneous magnetization gives rise to anomalous Hall effect (AHE) which scales with the net magnetization. The AHE is therefore not expected in antiferromagnets with vanishing small magnetization. However, large AHE was recently observed in certain antiferromagnets with noncolinear spin structure and nonvanishing Berry curvature. Here, we report another origin of AHE in a layered antiferromagnet EuAl2Si2, namely the domain wall (DW) skew scattering with Weyl points near the Fermi level, in experiments for the first time. Interestingly, the DWs form a unique periodic stripe structure with controllable periodicity by external magnetic field, which decreases nearly monotonically from 975 nm at 0 T to 232 nm at 4 T. Electrons incident on DW with topological bound states experience strong asymmetric scattering, leading to a giant AHE, with the DW Hall conductivity (DWHC) at 2 K and 1.2 T reaching a record value of ~ 1,5100 S cm-1 among bulk systems and being two orders of magnitude larger than the intrinsic anomalous Hall conductivity. The observation not only sets a new paradigm for exploration of large anomalous Hall effect, but also provides potential applications in spintronic devices.

cond-mat.str-el↗

arXiv2024

HLM-Cite: Hybrid Language Model Workflow for Text-based Scientific Citation Prediction

Citation networks are critical in modern science, and predicting which previous papers (candidates) will a new paper (query) cite is a critical problem. However, the roles of a paper's citations vary significantly, ranging from foundational knowledge basis to superficial contexts. Distinguishing these roles requires a deeper understanding of the logical relationships among papers, beyond simple edges in citation networks. The emergence of LLMs with textual reasoning capabilities offers new possibilities for discerning these relationships, but there are two major challenges. First, in practice, a new paper may select its citations from gigantic existing papers, where the texts exceed the context length of LLMs. Second, logical relationships between papers are implicit, and directly prompting an LLM to predict citations may result in surface-level textual similarities rather than the deeper logical reasoning. In this paper, we introduce the novel concept of core citation, which identifies the critical references that go beyond superficial mentions. Thereby, we elevate the citation prediction task from a simple binary classification to distinguishing core citations from both superficial citations and non-citations. To address this, we propose $\textbf{HLM-Cite}$, a $\textbf{H}$ybrid $\textbf{L}$anguage $\textbf{M}$odel workflow for citation prediction, which combines embedding and generative LMs. We design a curriculum finetune procedure to adapt a pretrained text embedding model to coarsely retrieve high-likelihood core citations from vast candidates and then design an LLM agentic workflow to rank the retrieved papers through one-shot reasoning, revealing the implicit relationships among papers. With the pipeline, we can scale the candidate sets to 100K papers. We evaluate HLM-Cite across 19 scientific fields, demonstrating a 17.6% performance improvement comparing SOTA methods.

cs.DL↗

arXiv2024

Magnetization dependent anisotropic topological properties in EuCuP

The correlation between magnetism and nontrivial topological band structure serves as a unique venue for discovering exotic topological properties. Combining magnetotransport measurements and first-principles calculations, we unveil herein that the hexagonal EuCuP holds topologically trivial state in the paramagnetic structure, while strong magnetization dependent anisotropic topological states in the spin-polarization structures. Specifically, it hosts a trivial topological state in the in-plane spin-polarization structure, while a Weyl semimetal state in the out-of-plane spin-polarization structure. Our scaling analysis suggests that the intrinsic Berry curvature in the spin-polarization structures can account for the observed large anisotropic anomalous Hall effect. First-principles calculations show that the magnetization and the spin-orbit coupling simultaneously play essential roles for the appearance of the four pairs of Weyl points in the out-of-plane spin-polarization structure. Our work therefore establishes in EuCuP the intimate relation between magnetism and the nontrivial topological states, which would be instructive for future study on this key issue of topological physics.

cond-mat.str-el↗

arXiv2024

Direct visualization of the impurity occupancy roadmap in Ni-substituted van der Waals ferromagnet Fe3GaTe2

Impurity substitution is a general strategy to study the intrinsic properties of a quantum material. However, when the target element has more than one Wyckoff position in the lattice, it is a big challenge but with extreme necessity to know the exact position and order of the occupancy of impurity atoms. Via comprehensive experimental and theoretical investigations, we establish herein the roadmap for Ni substitution in Fe3GaTe2, a van der Waals ferromagnet with the Curie temperature TC even reaching ~ 380 K. The results unambiguously reveal that in (Fe1-xNix)3GaTe2, Ni atoms initially form an van der Waals interlayer gap Ni3 sites when x < 0.1, and then gradually occupy the Fe2 sites. After replacing the Fe2 sites at x of ~ 0.75, they start to substitute for the Fe1 sites and eventually realize a full occupation at x = 1.0. Accordingly, TC and saturation magnetic moments of (Fe1-xNix)3GaTe2 both show nonlinear decrease, which is tightly tied to the Ni occupancy order as well as the different roles of Ni3, Fe1 and Fe2 sites in the spin Hamiltonian. The results not only yield fruitful insights into the essential roles of different Fe sites in producing the above room temperature high TC, but also set a paradigm for future impurity substitution study on other quantum materials.

cond-mat.str-el↗

arXiv2024

Estimating On-road Transportation Carbon Emissions from Open Data of Road Network and Origin-destination Flow Data

Accounting for over 20% of the total carbon emissions, the precise estimation of on-road transportation carbon emissions is crucial for carbon emission monitoring and efficient mitigation policy formulation. However, existing estimation methods typically depend on hard-to-collect individual statistics of vehicle miles traveled to calculate emissions, thereby suffering from high data collection difficulty. To relieve this issue by utilizing the strong pattern recognition of artificial intelligence, we incorporate two sources of open data representative of the transportation demand and capacity factors, the origin-destination (OD) flow data and the road network data, to build a hierarchical heterogeneous graph learning method for on-road carbon emission estimation (HENCE). Specifically, a hierarchical graph consisting of the road network level, community level, and region level is constructed to model the multi-scale road network-based connectivity and travel connection between spatial areas. Heterogeneous graphs consisting of OD links and spatial links are further built at both the community level and region level to capture the intrinsic interactions between travel demand and road network accessibility. Extensive experiments on two large-scale real-world datasets demonstrate HENCE's effectiveness and superiority with R-squared exceeding 0.75 and outperforming baselines by 9.60% on average, validating its success in pioneering the use of artificial intelligence to empower carbon emission management and sustainability development. The implementation codes are available at this link: https://github.com/tsinghua-fib-lab/HENCE.

cs.LG↗

arXiv2024

DomainStudio: Fine-Tuning Diffusion Models for Domain-Driven Image Generation using Limited Data

Denoising diffusion probabilistic models (DDPMs) have been proven capable of synthesizing high-quality images with remarkable diversity when trained on large amounts of data. Typical diffusion models and modern large-scale conditional generative models like text-to-image generative models are vulnerable to overfitting when fine-tuned on extremely limited data. Existing works have explored subject-driven generation using a reference set containing a few images. However, few prior works explore DDPM-based domain-driven generation, which aims to learn the common features of target domains while maintaining diversity. This paper proposes a novel DomainStudio approach to adapt DDPMs pre-trained on large-scale source datasets to target domains using limited data. It is designed to keep the diversity of subjects provided by source domains and get high-quality and diverse adapted samples in target domains. We propose to keep the relative distances between adapted samples to achieve considerable generation diversity. In addition, we further enhance the learning of high-frequency details for better generation quality. Our approach is compatible with both unconditional and conditional diffusion models. This work makes the first attempt to realize unconditional few-shot image generation with diffusion models, achieving better quality and greater diversity than current state-of-the-art GAN-based approaches. Moreover, this work also significantly relieves overfitting for conditional generation and realizes high-quality domain-driven generation, further expanding the applicable scenarios of modern large-scale text-to-image models.

cs.CV↗

arXiv2024

Fovea Transformer: Efficient Long-Context Modeling with Structured Fine-to-Coarse Attention

The quadratic complexity of self-attention in Transformers has hindered the processing of long text. To alleviate this problem, previous works have proposed to sparsify the attention matrix, taking advantage of the observation that crucial information about a token can be derived from its neighbors. These methods typically combine one or another form of local attention and global attention. Such combinations introduce abrupt changes in contextual granularity when going from local to global, which may be undesirable. We believe that a smoother transition could potentially enhance model's ability to capture long-context dependencies. In this study, we introduce Fovea Transformer, a long-context focused transformer that addresses the challenges of capturing global dependencies while maintaining computational efficiency. To achieve this, we construct a multi-scale tree from the input sequence, and use representations of context tokens with a progressively coarser granularity in the tree, as their distance to the query token increases. We evaluate our model on three long-context summarization tasks\footnote{Our code is publicly available at: \textit{https://github.com/ZiweiHe/Fovea-Transformer}}. It achieves state-of-the-art performance on two of them, and competitive results on the third with mixed improvement and setback of the evaluation metrics.

cs.CL↗

arXiv2023

Few-shot Image Generation via Masked Discrimination

Few-shot image generation aims to generate images of high quality and great diversity with limited data. However, it is difficult for modern GANs to avoid overfitting when trained on only a few images. The discriminator can easily remember all the training samples and guide the generator to replicate them, leading to severe diversity degradation. Several methods have been proposed to relieve overfitting by adapting GANs pre-trained on large source domains to target domains using limited real samples. This work presents a novel approach to realize few-shot GAN adaptation via masked discrimination. Random masks are applied to features extracted by the discriminator from input images. We aim to encourage the discriminator to judge various images which share partially common features with training samples as realistic. Correspondingly, the generator is guided to generate diverse images instead of replicating training samples. In addition, we employ a cross-domain consistency loss for the discriminator to keep relative distances between generated samples in its feature space. It strengthens global image discrimination and guides adapted GANs to preserve more information learned from source domains for higher image quality. The effectiveness of our approach is demonstrated both qualitatively and quantitatively with higher quality and greater diversity on a series of few-shot image generation tasks than prior methods.

cs.CV↗

arXiv2023

Absence of metallicity and bias-dependent resistivity in low-carrier-density EuCd2As2

EuCd2As2 was theoretically predicted to be a minimal model of Weyl semimetals with a single pair of Weyl points in the ferromagnet state. However, the heavily p-doped EuCd2As2 crystals in previous experiments prevent direct identification of the semimetal hypothesis. Here we present a comprehensive magneto-transport study of high-quality EuCd2As2 crystals with ultralow bulk carrier density (10^13 cm-3). In contrast to the general expectation of a Weyl semimetal phase, EuCd2As2 shows insulating behavior in both antiferromagnetic and ferromagnetic states as well as surface-dominated conduction from band bending. Moreover, the application of a dc bias current can dramatically modulate the resistance by over one order of magnitude, and induce a periodic resistance oscillation due to the geometric resonance. Such nonlinear transport results from the highly nonequilibrium state induced by electrical field near the band edge. Our results suggest an insulating phase in EuCd2As2 and put a strong constraint on the underlying mechanism of anomalous transport properties in this system.

cond-mat.mes-hall↗

arXiv2023

Direct observation of topological surface states in the layered kagome lattice with broken time-reversal symmetry

Magnetic topological quantum materials display a diverse range of fascinating physical properties which arise from their intrinsic magnetism and the breaking of time-reversal symmetry. However, so far, few examples of intrinsic magnetic topological materials have been confirmed experimentally, which significantly hinder our comprehensive understanding of the abundant physical properties in this system. The kagome lattices, which host diversity of electronic structure signatures such as Dirac nodes, flat bands, and saddle points, provide an alternative and promising platform for in-depth investigations into correlations and band topology. In this article, drawing inspiration from the stacking configuration of MnBi$_2$Te$_4$, we conceive and then synthesize a high-quality single crystal EuTi$_3$Bi$_4$, which is a unique natural heterostructure consisting of both topological kagome layers and magnetic interlayers. We investigate the electronic structure of EuTi$_3$Bi$_4$ and uncover distinct features of anisotropic multiple Van Hove singularitie (VHS) that might prevent Fermi surface nesting, leading to the absence of a charge density wave (CDW). In addition, we identify the topological nontrivial surface states that serve as connections between different saddle bands in the vicinity of the Fermi level. Combined with calculations, we establish that, the effective time-reversal symmetry S=$θ$$τ_{1/2}$ play a crucial role in the antiferromagnetic ground state of EuTi$_3$Bi$_4$, which ensures the stability of the topological surface states and gives rise to their intriguing topological nature. Therefore, EuTi$_3$Bi$_4$ offers the rare opportunity to investigate correlated topological states in magnetic kagome materials.

cond-mat.str-el↗

arXiv2023

Few-shot 3D Shape Generation

Realistic and diverse 3D shape generation is helpful for a wide variety of applications such as virtual reality, gaming, and animation. Modern generative models, such as GANs and diffusion models, learn from large-scale datasets and generate new samples following similar data distributions. However, when training data is limited, deep neural generative networks overfit and tend to replicate training samples. Prior works focus on few-shot image generation to produce high-quality and diverse results using a few target images. Unfortunately, abundant 3D shape data is typically hard to obtain as well. In this work, we make the first attempt to realize few-shot 3D shape generation by adapting generative models pre-trained on large source domains to target domains using limited data. To relieve overfitting and keep considerable diversity, we propose to maintain the probability distributions of the pairwise relative distances between adapted samples at feature-level and shape-level during domain adaptation. Our approach only needs the silhouettes of few-shot target samples as training data to learn target geometry distributions and achieve generated shapes with diverse topology and textures. Moreover, we introduce several metrics to evaluate the quality and diversity of few-shot 3D shape generation. The effectiveness of our approach is demonstrated qualitatively and quantitatively under a series of few-shot 3D shape adaptation setups.

cs.CV↗